Re: [vile] vile-9.6j.patch.gz

[Thomas Dickey]

On Wed, 26 Mar 2008, Thomas Dickey wrote:

> On Wed, 26 Mar 2008, Chris G wrote:
>
> > I don't quite follow this.  Maybe I didn't explain my case right.  It
> > should be impossible (as I understand it) to create a file with
> > iso-8859-1 pound signs in it if everything is set up correctly for
> > utf-8 shouldn't it?  I just happened to have some iso-8859-1 pounds in
> > a file I had created earlier.
> >
> > When I edited the file with the iso-8859-1 pound signs (just 0xA3
> > bytes by themselves) I got the incorrect cursor positioning. These are
> > invalid characters in utf-8 and, I believe, vile should really display
> > the 'invalid character' glyph which on my system appears to be an
> > inverse question mark.
>
> I see - can reproduce it here.  It looks as if vile's got confused at some 
> point (there are a half-dozen knots in display.c which I thought were all 
> aligned). It should show as \?A3 and track the cursor according to that 
> 4-column width.  The cursor is moving properly, but the display isn't.

This seems to fix it:

--- display.c   2008/03/06 01:10:52     1.471
+++ display.c   2008/03/26 23:06:14
@@ -732,7 +732,7 @@
     if (w_val(wp, WMDLIST)) {
        rc = vtlistc(wp, src, limit);
     } else if (b_is_utfXX(wp->w_bufp)
-              && column_sizes(wp, src, limit, &rc) == COLS_UTF8) {
+              && column_sizes(wp, src, limit, &rc) >= COLS_8BIT) {
        rc = vtlistc(wp, src, limit);
     } else {
        rc = vtputc(wp, src, limit);

Going down the third path, in vtputc it decides that the character is
really printable (since that would be true of the Unicode value, but
not the raw byte), and just stuffs it in the array that holds Unicode
values.

vtlistc is where it handles most characters other than the ASCII codes.
If it weren't in UTF-8 mode, then it might display that as \xA3.

--
Thomas E. Dickey
http://invisible-island.net
ftp://invisible-island.net